Me Desrochers : 819 743-2955  |  Me Roy : 819 208-6022 info@virtulexavocats.com

An Elections Canada worker gets the last three characters of a postal code wrong. One mail-in ballot comes back to sender. The election had been decided by a single vote. The Supreme Court of Canada has ruled: the election is annulled.

Some cases hang by a thread. This one hangs on one vote.

On 18 September 2026, the Supreme Court of Canada released its reasons in Sinclair-Desgagné c. Canada (Directeur général des élections), 2026 SCC 31. The disposition itself had been delivered from the bench on 13 February 2026: the federal general election held on 28 April 2025 in the riding of Terrebonne was annulled. Three of nine judges dissented.

For a litigation practice, the case is fascinating. Not because it is political — the Court carefully avoids that ground — but because it poses a question every citizen understands: when the state makes a mistake and knows it, whose job is it to fix it?

The facts: a label, a sticker, a lost vote

Toward the end of March 2025, an Elections Canada worker was preparing the address labels for the return envelopes accompanying mail-in voting kits in Terrebonne. The street number, street name, city, province and first three characters of the postal code were correct. Only the last three were wrong — the evidence showed he had inadvertently entered the last three characters of his own postal code. The four or so other volunteers working on the labels did not catch it either.

By the worker’s own estimate, at least 40 electors received a faulty envelope.

On 6 April 2025 — more than three weeks before polling day — he discovered his error. He destroyed some defective kits that had not yet been mailed. But he did not tell his supervisors and made no attempt to contact the electors already served. He told himself the error “did not appear to be causing any problem”: the returning office was receiving completed special ballots at a normal rate, including in envelopes bearing the wrong postal code. The returning officer for Terrebonne would learn of the error only after the election, when the media began asking questions.

One elector, Ms. Emmanuelle Bossé, had mailed her ballot 22 days before polling day, in the prepaid, pre-addressed envelope Elections Canada had provided — the very envelope the agency told her to use. Canada Post returned it with a sticker: “MOVED OR UNKNOWN […] RETURN TO SENDER.” Her vote was never counted.

After a judicial recount, Ms. Tatiana Auguste (Liberal Party of Canada) was declared elected with a majority of one single vote over Ms. Nathalie Sinclair-Desgagné (Bloc Québécois).

Contesting an election: a rare and demanding mechanism

The Canada Elections Act allows any elector or candidate to contest an election before the competent court on the ground of “irregularity, fraud, corrupt practice or illegal practice that affected the result of the election” (s. 524(1)(b)). If the grounds are established, the court “may annul the election” (s. 531(2)). For federal matters in Québec, the Superior Court hears the application, and the appeal lies as of right to the Supreme Court of Canada.

The Act does not define “irregularity.” It was Opitz v. Wrzesnewskyj (2012) that set the framework, in three steps:

  1. The applicant must prove there was an irregularity — which includes a serious administrative error;
  2. The applicant must prove the irregularity affected the result of the election by affecting the vote count;
  3. The court then decides whether to exercise its discretion to annul the election.

The burden rests on the applicant at each step, on a balance of probabilities. The Superior Court (Dufour J.) had dismissed the application at the first step: the postal code error was simple human error, made without dishonest or malicious intent, and therefore not an irregularity. The judge added that even had he found an irregularity, he would not have annulled the election.

Step one: an honest mistake can be an irregularity

This is where the decision makes its doctrinal contribution. Jamal J., for the majority, holds that the trial judge erred in law by requiring that an administrative error be intentional or made in bad faith.

“Unintentional, honest or good faith errors can also taint the integrity of the electoral process if they prevent electors from exercising their right to vote,” he writes. The key consideration is not intent, but whether the administrative error was capable of undermining an elector’s right to vote.

The majority points out that in Opitz itself the Court had characterized as an irregularity a situation “consistent with a clerical error”: a deputy returning officer had signed voter registration certificates where the electors were required to sign.

A striking point: the Court accepts that no specific provision of the Canada Elections Act was breached. Mail-in voting, though common since the COVID-19 pandemic, is not expressly provided for in the Act; the precise procedure is set by the Chief Electoral Officer as an administrative practice. But the absence of a statutory breach does not prevent a finding of irregularity.

What tips the analysis is the combination of two elements: the typographical error, and Elections Canada’s failure to take any corrective step during the three weeks before polling day, when the agency had the affected electors’ contact information and its own Special Voting Rules administrator testified that she would have expected the worker to report the error.

Step two: who bears the risk of mail-in voting?

The elected member’s main argument rested on two provisions: s. 239(2) and s. 240 of the Act, which provide that an elector “shall ensure that” his or her special ballot is received at the right place at the right time, and that the elector “has the sole responsibility” to do so. Ms. Bossé, it was said, had a tracking number; she could have checked, discovered the problem, and voted in person.

The majority rejects that reading. Tracing the legislative history back to the 1993 reform and the Lortie Report, it concludes that the purpose of these provisions is to promote accessibility while requiring diligence from electors in the face of risks outside Elections Canada’s control: postal delays, an elector who mails too late, similar external risks. They are not there to relieve the agency of the consequences of its own errors, which it has the power to correct.

Reading them otherwise would, in the Court’s view, produce absurd results: Elections Canada would never bear legal responsibility for its own failures in the special ballot system, so long as an elector theoretically had another way to vote.

The Court adds an important — and slightly counterintuitive — clarification: the trial judge was wrong to take into account the fact that Ms. Bossé had voted for the second-place candidate. Secret balloting bars any analysis of how electors actually voted, and that bar applies at every step of the analysis.

Step three: the “magic number” test

At the final step, Opitz supplies a tool of brutal simplicity: the “magic number” test. If the number of votes affected by the irregularity equals or exceeds the margin of victory, doubt remains as to the true winner, and it would be unreasonable not to annul the election.

Here, one affected vote, one vote of majority. The numbers match exactly. The Court notes that the test inherently favours the applicant, since it assumes the affected votes would have changed the outcome — an assumption that may be highly improbable — but that no substitute test has yet been developed.

The Court sets aside the three reasons given at first instance for refusing annulment: the status of three ballots contested at the judicial recount (a matter the Act expressly excludes from a contested-election application), the means available to Ms. Bossé to track her ballot (it was not for her to make up for Elections Canada’s error), and the fact that she could have voted in person (“true, but beside the point”).

The majority is careful to reassure: the irregularity found is narrowly circumscribed and will not open the floodgates. In many cases, the margin of victory will be large enough to prevent the magic number test from being met.

The dissent: three judges would have left the result standing

Karakatsanis and Martin JJ., with Moreau J. concurring, would have dismissed the appeal. Their reasoning deserves attention; it is anything but marginal.

In their view, mail-in voting is a distinct regime, voluntarily chosen by the elector, carrying known and foreseeable risks at every stage. Parliament responded to those risks precisely by placing on the elector the responsibility to ensure the ballot arrives — and nothing in the text, history or purpose of the provisions limits that responsibility to errors external to Elections Canada. “The source of an error […] is irrelevant to the application of these provisions.”

On seriousness, they recall that “irregularity” sits beside “fraud, corrupt practice or illegal practice”: the conduct must be comparable in nature and gravity. A mistyped postal code flowing from an honest error does not reach that level. They cite a 2013 Elections Canada report finding non-compliance with administrative rules in 1.3 percent of all voting transactions. “Humans are fallible. And if we want to have confidence in the finality and legitimacy of our elections, then we must accept a small measure of human error.”

Their warning is direct: if this error is enough to reopen a result, “it is difficult to see what type of error would not meet the threshold.” Such a low bar would put every election result at risk.

Why this case stands out

First, for the rarity of the remedy. Annulling an election disenfranchises not only the wronged elector but everyone who voted in the riding. A by-election is never a do-over: political circumstances shift, information changes, some voters tire. The majority acknowledges that a by-election is “not a perfect solution”; the dissenting judges go further, noting that circumstances evolve, that electors come to be inconvenienced and that some become disillusioned.

Second, because the case moves a boundary: it establishes that a public administration can see a result annulled for a serious administrative error that it discovered itself, that it had the ability to correct, and that it did not correct. That is not punishment; it is consequence. And it is a principle that reaches well beyond electoral law.

Third, because what followed vindicates the dissent in a practical sense. A by-election was held in Terrebonne on 13 April 2026. Ms. Auguste won again, this time by several hundred votes. The doubt about the true winner, which legally justified annulment, was resolved by the voters themselves — two months after the annulment.

That does not undercut the majority: the legitimacy of a result is not measured in hindsight. But it illustrates the genuine tension the Court had to resolve between the integrity of the process and the finality of results.

Key takeaways

If you vote by mail, do it early and use the tracking number Elections Canada provides. The Supreme Court confirms that a diligent elector who follows official instructions need not second-guess the documents the state hands her — but it also confirms that responsibility for getting the ballot delivered rests first with the elector for everything outside the agency’s control.

If you are a candidate or organizer, remember that the burden is heavy — and that an appeal to the Supreme Court must be brought within eight days of the decision. A contestation must rest on concrete proof of a specific error, a direct causal link to uncounted votes, and a sufficiently narrow margin of victory. The rigour of the test makes annulment an extremely rare outcome.

If you run any kind of process — electoral, professional, associative or union — the lesson transposes, and it is a stern one. Discovering an error and not reporting it because it does not appear to be causing a problem is precisely the conduct the Supreme Court sanctions. The duty to correct arises from knowing about the error and being able to fix it, not from how serious it looks at the time.

And on principle, the decision recalls something easily forgotten: the right to vote guaranteed by section 3 of the Canadian Charter of Rights and Freedoms is not subject to the notwithstanding clause. It is not an administrative privilege. Every vote counts — and here, literally.

Related reading


This column is published for general information purposes only and does not constitute legal advice. Because every situation is unique, you should consult a lawyer for advice tailored to your own case. Feel free to contact Virtulex avocats to discuss your situation.

William Desrochers, Attorney, Virtulex avocats — www.virtulexavocats.com

Sinclair-Desgagné c. Canada (Directeur général des élections), 2026 SCC 31 (Supreme Court of Canada, file 42076; heard and judgment rendered 13 February 2026, reasons filed 18 September 2026; reasons of Jamal J., Karakatsanis, Martin and Moreau JJ. dissenting) — read the decision on the Supreme Court of Canada website